Brittle fracture during fatigue loading at -10 degrees c of a full scale tubular X-joint

1.1W. doc. X1-093-85

Deift University

_{of Technology}

DEPARTMENT OF MARINE _TECHNOLOGY

SHIP STRUCTURES LABORATORY

BRITTLE FRACTURE DURING FATIGUE LOADING

AT _100 C OF A FULL-SCALE TUBUlAR X-JOINT

BY

PROF. IR J. J. W. NIBBERING

IR. REN ZIJIN

JUNE 1985

Deift University of Technology

II.W. doc.

X1-09-B5

DEPARTMENT OF MARINE TECHNOLOGY

SHIP STRUCTURES LABORATORY

REPORT NO. r:: SSL 289

1. Introduction

In offshore structures the use of steels with high yield point is often promoted by designers and steel makers. On the other hand the fabriçaters of welding materials are more reluctant. They realize too well that weld defects are not easily avoided and are more dangerous in high strength

steel structures. Furthermore the strength and toughness of the base metal may be seriously impaired in weld regions. Apart from that, the welding - including pre- and post-heating - is expensive, notwithstanding

the fact that the amount of work is reduced due to the smaller thickness-es. For cyclically loaded structures post-weld heat treatment is an ab-solute necessity in order to reduce the high welding stresses and remove hydrogen.

In general the authors are not so afraid of welding stressés in sea-loaded structures made of lower strength steels. Incidental high loads will free the structure soon from them, at least at the points of highest

stresses.

But in high yield structures the ratio between load stresses and yield point is smaller than in lower yield structures, so that residual stress-es do not relax as soon and as much as in the latter. This meansthe pos-sibility of quicker initiation of cracks (even if the stress level is the

same in both structures!) nd quicker propagation of cracks, as a

con-sequence of absence of crack closure under compressive loads.

The main reasons for using high strength steels for instance in jack-up

legs are that cklingstrenthis increased and of course weight is

re-duced. But then the problem how much the compressive stability is impair-ed by the presence of eventual cracks becomes more important. But this is outside the scope of this paper.

Here attention will be focussed on the fatigue behaviour at -10°C, which resulted in a really alarming case of brittle fractùre.

Fatigue loading at low temperature has been chosen for this experiment, because it conforms best to actual situations. An even more important reason is that fatigue cracks travel through the structure, thus. pro-viding a large amount of possible initiation points for brittle fracture. When the crack tip meets a spot where the quality of base metal, weld or H.A.Z. is too low a brittle fracture may develop /1/, /2/.

_____7: :

A

rig for

"-4OO tons

copresson

r

-4-2. Test specimen

The specimen was a full scale tubular X-joint consisting of tubes of 368 X 20 mm. It was made of steel with a yield point of 850 N/mm2 and a tensile strength of 925 N/mm2. The chemical composition was 0.13 C; 0.34 Si; 1.1 Mn; 0.02 P; 0.015 S; 0.06 Al; 1.18 Ni; 0.24 Mo; 0.08 V; 0.51 Cr; 0.012 N. For Charpy properties, see section 6.

The weld metal properties were:

Chemical C M Si Mo Ni Cr

composition 0.07 1.4 0.3 0.4 0.9 0.2

Impact (temp.) -20 -40 -20 -40

energy J 135 80 140 80

The welding and N.D.T. were carried out with the utmost care according to usual modern practices.

Figure 1 shows the specimen in the test rig. The special topic in this

case was that the two tubes were not mutually perpendicular but position-ed at an angle of 60°. The member, which is mainly under compression in the actual structure was the continuous one.

3. Test set-up (fig. 1)

The specimen had been built in in the 1000 tons tension-compression fatigue testing machine of the Deift Ship Structures Laboratory.

A separate rig had been constructed in order to be able to keep the load

on the continuous tube (transverse in the machine) in static compression

at -400 tons, (fig. 2).

The interrupted tube was loaded cyclically by the machine. The specimen

Yielding Non-welded Stress relieved

strength N/mm 620 600

Elongation

21 21

at fracture

Name Conarc 70 (basis layer) Conarc 85

Code ASME SFA-5.5 E100 16G ASME SFA-5.5 E120 12G

6

had been provided with a large number of strain gauges in order to be able to measure the strains at hot spots, where cracks may start.

During the fatigue loading the output of some gauges was constantly

recorded, in order' to be able to trace the first moments of crack init-iation and follow the crack propagation.

4. Fátigue test

The fatigue test was divided into two parts.

Fatiue load with ratio R = -1 Actual fatigue load:

F = +300 ton on the cut ttibe.

max

F . -300 ton axial stress is 19.9% of the yield.

Frequency = 0.033 cycles/sec.

P = -400 tôn on straight tube, axiai stress is 26.77. ,f the yield. Temperature = 15°C.

Fatigue cycles = 1100.

Crack initiation:

After 1100 fatigue cycles,, three cracks were found at three toes of the weld where peak strains occurred. The depth of the cracks was

very shallow (about I to 2 mm). These cracks could only be seen with

a magnifying glass under a tension load more than 275 ton. The places where cracks occur are shown in fig.. 3.

Fatigue load with ratio R = -0.5 Actual fatigue load:

F = +400 ton on the cut tube.

max

F . = -200 ton.

mn

Frequency = 0.037 cycles/sec.. P = -400 ton on the cut tube.

Fatigue cycles = 1068 under temperature = -5°C.

Fatigue cycles = 116 under temperature = -10°C.

Crack propa:ation: .

From inspection of the cracks thé crack propagation process could be

'reconstructed: .

1) crack with a length of 115 mm (crack I).

This crack first propagated along the toe of the weld in the wall of the continuous tube. Then one tip of this crack left the toe

(Fotocode : A) Crack U: 30mm (underside) Crack 1:115 mm(upperside) Fracture I (upperside)

A1/B1

I-6

Fig. 3

Situation of cracks (view from top side)

/ B4

n-o

U -12

Fracture U (underside)

Crack : 150mm after fracture (underside)

)Crack

: 18mm (upperside)

8

of the weld in the neighbourhood of gauge 164, and propagated into the longitudinal direction of the continuous tube. It stopped at a point about 80 mm from the toe. Another one left also the toe of the weld and propagated as a brittle fracture in the longitudinal direction of the continuous tube.

Crack with a length of 30 mm (crack II).

This crack did not propagate through the wall, which means that the propagation in depth was small.

Crack with a length of 18 mm (crack III). This cracl did not propagate through the wall. Crack IV.

This crack had not been detected after the first fatigue stage. It propagated completely through the Continuous tube in a brittle way.

5. Discussion of the fatigue results

Figure 4 shows results of Dutch experiments /3/, /4/ by Dijkstra and De Back et al. with tubular joints of various diameters. 'The well known

influence of scale appears clearly. In the figure the AWSXX curve is also given together with one point of a new British fatigue guidance note for 32mm thick tubular joints, (mean less two standard deviations).

The present result is in 'line wici the curve for ₀ 457 for steel with

yield point 350 N/mm2; (the curve for ₀ 368 was estimated.)

It would be of interest to know whether the transverse constant compress-ive load of 400 tons has perhaps had a negatcompress-ive influence on 'the fatigue behaviour. At this moment no answer can be given, but it should be real-ized that 400 tons conforms to a stress of only 20%a .(in. /4/ no influence was found). However it may be, it is quite obvious that the fatigue strength

Of joints made of high strength steels is _{not superior to that of joints}

made of lower strength steels. For the high stress region this is surpri-sing as one might expect especially there eventual benefits from the'use

of high strength material. The _{present test result refers. to very shallow}

cracks (about 2 mm. deep) because a brittle fracture occurred at that

stage of the test. In the case that the experiment would have been carried out at + 20° C instead of - 10° C the fatigue test would certainly have

continued another 2000 cycles leading to a more favorable point in the

Present test high yield

L

860 N/mm2

*

5500 5000 e C 1000 500 300 10

--'

_IO' 1.2- Z.0x103 cycles 5280

Extension proposed by fida 151

Estimated

:

Brit. guide 32mm

-

rnean-2S.D. O D

'N.Ø168

a .

48 R

0457

..

CC lO

riumbu et cyctel (end et tcst)

Fig.4

Dutch test results on T- and Xjoints of circular hollow sections. (31, (4]

Fig. 7. _{Brittle fractures in transverse tube.}

Detti UniwriIy et Tcchnology/ T. ti. O.

o T-ø6B 3.O.5 Rio Ax/tP

A T-0457 3.Q5 Rio Ax

C-LS7 (3.1.0 R-4 Ax a

-94 3O.5 Rio Ax

+ X-0914 .O.5 Rio Ax

-9-o a o

After fracturing of the specimen inspection of the crack surface revealed that there were several points where cracks had started (fig. 5, 6).

This. is of interest because it shows that the _{fatigue performance was}

not caused by a single,, particularly dangerous defect.' On the contrary all cracks started - as usual in defect-free welds - at the fusioñ line - of the welds. Therefore the fatigue result of the specimen should be

con-- sidered as rather characteristic for this material in the longitudinal dicon-- di-rection (crack path).

For, not any fatigue crack has initiated alongside the welds in the inter-rupted tube (longitudinally in the machine).

Apparently cracks did not develop as easily transverse to the tube axis as they did in line with it; the resistance to crack initiation seems to be better. But it should be acknowledged that the stresses at the fusion line of the interrupted tube were at least 20Z lower than those at the fusion

line of the transverse continuous tube /6/.

Some. infòrmátion about the crack propagation at high nominal stresses was

obtained from the fatigue results of the fabricated COD-specimens.

For a K-value of 1440 N/mm3"2 da/dN was about3,5x1O for cracks

develop-ing in the longitudinal tube material and about 2.5 X 1.O for

trans-verse cracks. These values are about half as high as those for steel Fe 510.

12

-6. Brittle fracture of the specimen at -10°C

In section 3 the test procedure has been described.

The latter 1184 of a total of 2284 load cycles had been applied at low temperáture. Of these 1068 occurred at -5°C, the latter 116 at -10°C. The fatigue load vas +400 ton/-200 ton, corresponding to nominal stresses

0.27 a

/0.13

a

y

y

The test ended with a spontaneous brittle fracture. Figures 5 and 6 show

the crack paths on both sides of the specimen. Figure 7 shows how far

the brittle fracture extended in the transverse tubes. This is astonish-ing bearastonish-ing in mind that in the greater part of the transverse tubes hardly any stresses were present perpendicular to the crack path.

The starting point(s) of brittle fracture can be seen in figures 8 and

9. The fracture probably did not develop in one moment but in stages.

This can be seen in fig. 8.

It seems that at least four partial fractures developed prior, to the final one.

The strain gauge records confirm this more or less. Of six strain gauges a continuous record was made during the fatigue loading at low temper-ature. But unfortunately they were not situated close (enough) to the part of the structure shown in fig. 8.

The gauge near point A II-4 in fig. 8 was connected to a peak detector (double amplitude of strain) which became active once in 12-13 load cycles. The result is shown in fig. 10. (The number of cycles refers to

those effectuated at -5°C and -10°C). At 1000 there is a clear change in

slope which might be due to a sudden extension of the shallow rusty fatigue crack at the edge of the plate. A second change of slope (to three horizontal points) occurs at about 1130 cycles. Next a small shift of thé record is visible (two points), which again can only be caused by

rather abrupt crack extension. The starting point of the final fracture will have been either at the left or the right side in fig. 8.

It is of interest that the partial brittle fractures occurred mainly in

the welded region (weld and H.A.Z.) and not in the base metal of the

transverse tube. This confirms what will be discussed also in. the next

section that the resistance of the unwelded base material against

longit-udinal cracking was worse than that of the weld and H.A.Z.

There is a possibility that primary cracking has started at the other

side of the specimen (fig. 9), . because the surface was more. corroded

P) '-1 rP (D H- rP H-H- (D.. P) CO (Dr? H '1 H DCl) O

l H-

(D(D (D -S O

o H

(D

I-tij

'4 r? cI)C() (Dpl(D (D I-ti P) ti r H- (D Il Q H-Cl) r-t rP w P) _'... li (D 01-ti P) ti n _{P)rP H- pl (Dp) n} .. C/) r? II Cl)

14

-strain gauge in the vicinity of A 1-6 in fig. 11 supports this view.

The shear lips all over the length of the fracture surface were rather small (fig. 8 left and right). Only at the origin and at a few inter-mediate spots and of course at the arresting points the shear lips were

thicker. This confirms that the propagation of the brittle fracture oc-curred very 'easily'. It suggests that the material of the transverse tube would probably not have been able to arrest cracks at somewhat, higher temperatures, say 0°C. This is confirmed in the next section.

7. Comparison with Charpy and C.O.D. results; influence of ageing

a. Char2y tests

The material had been supplied in 1980. Then the Charpy energy at -40°C

was 44 J (notchJ_ tube axis). Lloyd's requirement was 2 J, so a wide

margin was present.

After the experiment described in this report new Charpy bars were

fabric-ated and tested (fig. 12). Now only 32 J was obtained at -40°C.. I.t is

curious that also the yield point and the tensile strength had changed during the course of 5 years:

Apparently the material is sensitive to ageing. This means :that structures made of that material .(andper.haps similar ones) are lesssafe nowadays

than at .the time of fabrication

Another alarming thing is that the Charpy value for bars with notches in the longitudinal diréction was as low as 22 Jat -40°C. At -10°C (the testing temperature of the full scale specimen) the energy was still only

28 J. . . .

In fig. 12 are. included requirements for similar material according to Euronorm 113-72 (Fe E 355-KT). The correspondence is good. Then it must be concluded that for offshore applications - especially for parts above

sea level - these requirements are not satisfactory. Figure 13. gives in-formation about the fracture surface of the Charpy bars. For the tube thickness concerned (20 mm) át least 50% shear .surface is necessary, when crack arrest capability of the material is desired. Even for the L-bars this will on.ly be possible at about 0°C. From the D- and T-results, which

1980 1985

a 810 875

y a

5 ....UOLT

4-,

A M

3-.

P L ¡ T

2_

U o E S ...VOLT

4_

A N

3.-P L I T

2_

U o E 1.-I I - _I 0 200 400 600 000 1000 1200 1400 1600 1000 2000

load cycles

Fig. 10

Strain amplitude close to A114. (fig.8)

a

ituatiri of

stra ngauge

1 /

r,ck Iv

- - . 0 200 400 500

000. 1000

1200 1400 1500 1000 2000

load cycles

Pig. li

_{Strain amplitude near A15 (fig.9)}

16

-are indicative for the full scale test result, it may be concluded that crack arresting will only be possible at temperatures abo.ve +20°C.

A point which needs consideration is whether the initiation of the brittle fracture was influenced or not by the quality of the weld and the heat-affected zone. In fig. 14 Charpy results for weld and H.A.Z. of a similar structure as tested, made by the same firm, are added to the values for

the tube material of fig. 12. It may be concluded that the poor result of

the full scale test was not due to inferior welding.

b. C.0.D.-testinZ

The C.0.D.-tests showed a clear difference in toughness for cracking in the transverse and in the longitudinal direction of the tubes.

For the transverse direction the CTOD at -16°C was equal to 0.7 imn and at -25°C and -40°C about 0.25 mm. For the longitudinal direction it was only 0.22 mm at -4°C and 0.28 nun at +39C.

This result supports what has been observed in the full scale test, where

no brittle fractures had developed in transverse sections of the

longit-o

udinal tube at -10 C.

8. Conclusions

The high stress - low cycle fatigue strength of tubular cross

connec-tions made of high strength steels is about equal to that for

lower strength steels.

Shallow cracks may develop into brittle fractures when thé fatigue loading is applied at moderately low temperatures.

These brittle fractures are not (easily) arrested when running in the longitudinal direction of the tubes.

Ageing of the material may aggravate seriously the situation for actual structures when they become older.

The notch toughness of tube material with respect to longitudinal cracking is a lot worse than that to transverse cracking.

Fatigue loading at low temperature proved to be once again a very realistic and efficient testing procedure for large structural

References

/1/ J.J.W. Nibbering and A.W. Lalleman:

'Low cycle fatigue tests at low temperature with E.C.-welded plates'. 11W-doc. X-593-70.

/2/ J.J.W. Nibbering and H.C. Scholte:

'Realistic testing of welds by fatigue bending at low temperature'. 11W-doc. X-1014-82. ISP, Vol. 31, March 1984, No. 355.

/3/ 0.D. Dijkstra:

'Fatigue strength of tubular T- and X-joints'. OTC 3696, May 1980.

/4/ J. de Back:

'Vermoeiingsgedrag van offshore s talen buisknooppunten'. Report for Annual Conference of SMOZ, Dec. 1984.

/5/ K. lida:

'Fatigue strength of welded tubular K-joints of 800 N/mm2 class high strength steel'.

11W-doc. XV41978.

/6/ Ren Zijin:

'Full scale test and finite element calculation for a tubular

H

X-joint'.

Deift, May 1985. Graduate work Delft University of Technology, Deift Ship Structures Laboratory.

17-

90-80

70

60

i Q

-

18-Fig. 12

O and T I...

/

r-1.2 e. Z 1.9 Euronorm 11372 FeE 355-KT Charpy-V (150-V) v

e--.-.

.

.L8 1.7 O

--60

-.50

-40

-30

-20

-10

0

10

+20

Test temperature (°C

4.-Q, C ai

>

I 30

>

D. I-.

LI

20

.1.6

0 1.5 1.15 T.16

,

/

/

/

.1.3

'e.

T.13

'

xT.16

0.8

0.7

0.2

-I

0.1 1.10/11

70

50

10

-60

Fig. 13

Euronorm 113-72

FeE 355-Kl

Charpy-V (lSD-y) LO 1.7 1.16 0.7/8,

XT.16

6

jis

I I _I J

-50

-40

-30

-20

-10

0

+10

+20

Test temperature (

ft)

1.2

f L.9

L.6

L.1

1.3 X T. 10/11

0.2

- 0.1

0.4 03/ T. 13

90

I-(n

u

-J

o

s

> 40

L. a,

C

w

>

30 >% Q. L. ru -C

L)

20 10

Fig. 14

-

20

-80

70

H.Z.

60

I

+ 0.6 -HZ. HZ. H.Z. w.Z. w.z.. w.z. o H.H. oH.H.

oftH.

1.7 W.H. 2xA W.H. tlW.H. 1.2 W.ft (weld heeL) H.H. (H.A.Z. heel) WZ. (weld on side) HZ. (HAL on side) Euronorm 113-72

FeE 355-Kl

Charpy-V (ISO-V)

- s

1.16 0.7 0.8 01

-60

-50

-40

-30

-20

-10

0 +10

+20

Test temperature (°C)

10765 L.6

I--fi..5

1c T. 15

/

,

,

,

L.3 1.13 e .

-.

D.1 02 T.10/11

Delf.t University of Technology 11W-doc. X-1093-85 Department- of Narine Technology

SHIP STRUCTURES LABORATORY

Report No.

-SSL 289a

ThRITTLE FRACTURE DURING FATIGUE LOADING

AT -1-0°C OF A FULL-SCALE TUBULAR X-JOINT

(revised version with new sect.ion on ageing)

by

Prof.ir. .J.J.W.Nibbering

Ir. Ren Zij-in

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

BRITTLE FRACTURE DURING. FATIGUE LOADING AT -10°C OF A

FULL-SCALE TUBULAR X-JOINT. INFLUENCE OF AGEING.

*

J.J.W. Nibbering and Ren Zijin

A full-scale cross: connection between two tubes has been subjected to high-stress low-cycle fatigue loading.at -10°C. A brittle fracture developed after a small number of load cycles. The yield point of

the material which originallywas equal to81ON/2,

had obtained a value of 885 N/2 after 5 years of rest. Therefore special attention has been paid to the ageing characteristics.

INTRODUCTION

Inoffshore structures the useof steels with high yield pOint is often promotedby designers and steel makers. On the other hand the fabricators Of welding materials are more reluctant. Th.ey realize too well that weld defectsare not easily avoided and that they are more dangerous inhigh strength steel struturèsthan--in structúres made of mild steel. Equally important is, that the strength and toughness of the base metal may be. seriously impaired in weld regions. The welding itself - including pre- and post-heating - may be expensive, despi:te the reduction in work

due to the smaller thicknesses. For cyclically loaded structures post-weld heat treatment isvery desirable. in order to reduce the

high welding stresses. and remove hydrogen. .

-In general the authors are not so afraid of welding stresses in sea-loaded structures made of lower strength steels. Incidental high loads will free the structure soon from them, at least at the points of highest stresses. But in high yield structures the ratio between load stresses and yield point is smaller than in lower

Department of Marine Technology, ShIp Structures Laboratory, Delf t University of Technology.

The specimen was a full scale tubulàr X-joint consisting of tubes 'of 368 X 20 mm. It vas made.of steel with aieid point of

850 N/mm2. and' a tensile.strength of' 925' N/mui . The chemical com-position was 013C.; 0.34 Si,; I.! Mn; 0.02 P; 0.015 S; 0.06 Al;

1.18.Ni; 0.24' Mo; 0.08 V; 0.51' Cr; 0.0.12. N. For Charpy properties,

see Figures 12,, 1.3 and 14.

The weld metal properties were: TABLE I

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

yield structures, so that. residual stresses do not relax as soon 'and as muchas in the latter. Th.is means the possibility of..quicker

initiation of cracks (even if the stress level is the same in both structures.!) ànd' quicker propagation of cracks,. as a consequence of absence of crack closure under'compressive loads.

The main reasons for using high strength steels for instance in j:ackup legs are that the.. compressive strength of the chords and braces is increased and consequently the weight is reduced But

then a new problem appears, viz.how much thecompressive. stability is impaired by the. presence of eventUal, cracks. But this is outside

the scope of this paper. Here attention Will be focussed on the fatigue behaviour at -10°C, and the .invo].ved.risk of brittle f

rac-ture.'Fatigue loading at low temperature. has been chosen for this experiment, because it cónforms bes:t'to'actualsituations.' Fatigue cracks may travel through the structure, thus providing constantly new possible initiation points for brittle fracture. When the crack tip..meets.a spot where the quality of base metal, weld or H.A.Z.' is low, a brittle fracture may develop (Nibbering and Lalleman (1), Nibbering. and Schoite (2)).

TEST SPECIMEN

.:Jtfle Conarc 70 (root passes) ' ' Coitare 85...

Chemical . C M Si Mo, Ni Cr

composition 0.07 1.4 0.3 0.4 '0.9 0.2

Yielding Non-welded ,. 'Stress relieved

strength N/mm .' 620 ' 600 Impact (temp.) -20 -40 ' -20 ' -4.0 energy J 135 80 140 80 161 .2' Elongation 2,1 21. at fracture

Code ASME SFA-5.5 E 100 16G . ASME SFA-5.5. E 12012G

FRACTURE CONTROL OF ENGINEERING STRUCTURES ECF 6

The welding and N.D.T. were carried out with the utmost care according to usual modern practices..

Figure .1 shows the specimen in the test rig. The special topic in this case was that the two tubes were not mutually perpendicular but positioned at an angle of 60 . The member, which is mainly under compression in the actual structure was the continuous one.

TEST SET-UP (FIGURE 1)

The specimen had been built in in the 1000 tons tension-compression fatigue testing machine of the Deif t Ship StrUctures Laboratory. A separate rig had been constructed in order to be able to keep the load on the continuous tube (transverse in the machine), in static compression at -400 tons, (Figure 2).

The interrupted tube was loaded cyclically by the machine. The specimen had been providéd with a large number of.strain gauges in order to 'be ablé to measure the strains at hot spots, where cracks may start. During the fatigue loading the output of some gauges was constantly recorded., in order to be able to trace the first moments of crack initiation and. follow the crack propagation.

FATIGUE TEST The fatigue teSt was divided into two parts. Fatigue Load. with Ratio R = -1

Actual fatigue load:

F +300 ton on the interrupted tube.

Fm,x = -300 ton axiál stress is 20% of. yield point.

min

Frequency = 0.033. cycles/sec.

= -400 ton on the continuous tube., axial stress is 26.7%-o! yield point.0

Temperature = 15 C. Fatigue cycles = 1100.

Crack. initiation. After I 1OOE fatigue cycles, three cracks were found at..three toes of welds where peak strains occurred. The depth of the'cracks 'was.very shallow (about ito 2'). These cracks could only be seen.with..a. magnifying glass under a tension.load above 275. ton. The places where cracks occurred are shown in Figure

3.

Fatigue Load with Ratio R = -0.5 Actual fatigue. load:

F - +400 ton

max

F.

-200 ton.

min

on the interrupted tube.

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

Frequency = 0.037 -cycles/sec.

P = -400 ton on the continuous tube.

Fatigue cycles = 1068 under temperature =

-5C.

Fatigue cycles = 116 under temperature = -10 C.

Crack propagation. From inspection of the cracks the propagation

process could be reconstructed:

Crack with a length of 115 mm (crack I).

This crack first propagated along the toe of the weld in the wall of the continuous tube. Then one tip of this crack left the toe of the weld in -the neighbourhood of gauge 164 and pro-pagated into the longitudinal direct-ion of the continuous tube. It stopped at a point.about 80 mm from the toe. Another one left also the toe of the we-id and propagated as a brittle fracture in

the longitudinal direction of the continuous tube. -Crack with a length -of 30 mm (crack II).

This crack did not propagate through the wail; in fact the pro-pagation in depth was small.

C-rack with a length of 18 mm (crack III).

This c-rack did not propagate through the wall.

-Crack IV.

This crack had not been detected after the first fatigue stage. It -propagated completely through the continuous tube in a

-brittle way.

- DISCUSSION OF THE FATIGUE RESULTS

Figure 4 shows- results of Dutch experiments by Dijkst-ra (3) and De Back et al. (4) with tubular joints of various diameters. The well known influence of -scale appears clearly. In the f igüre the AWS-XX curve is also given together with one point of a new British

fatigue -guidance note for 32 n thick tubular j-oints (mean less two

-standard deviations) The present result is in line with the curve

for ₀ 457 for steel with yield point 350 N/mm2 and with the one

estimated for ₀ 368. - -- -

-It would be of interest to know whether the transverse constant compressive load of -400 tons- has perhaps had a negative- - influence on the- fatigue .behaviour. At this moment no answer can -be given, but--it should be realized that 400 tons conforms to a stress of only 27Z ay. (in (4) no influence.was found). Howe'er it may be, it is confirmed once again that the fatigue -strength -of joints made of h-igh strength steels is not superior to that of, joints made of lower strength steels. That this is also t-rue for the high stress region is disappointing because- when benefità from the use -of high strength material are hoped for, it is in that -region. -That this -hope is not fulfilled is- due to the residual welding stresses.

The present -teat result refers to rather small cracks because a brittle fracture developed soon. In- the case -that the experiment would have been carried out at +20°C instead of -10°C. the fatigué

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

test could probably have been continued for ano.ther 2000 cycles leading to a more favourable point in the diagram.

After fracturing of thespecimen inspection of the crack sur-face revealed that there were several points where cracks had started (Figures 5 and 6). This is of interest because it shows that the low fatigue result was not caused by a single, incidentai-ly serious defect. For all cracks started - as usual in defect-free welds - at the fusion line of the welds. Therefore the fatigue result of the specimen may be considered 'as rather characteristic for. this material in its longitudinal direction (crack path). Not àny fatigue 'crack has initiated at the fusion line on the side of the welds in the interrupted tube. Apparentlycracks did' not

develop. as 'easily transverse to the tube axis as they did in line with it,; the resistance to crack initiation seems to be better. But it should be acknowledged that the stresses at the fusion line of the interrupted tube were some 20% lower than those at the fusiön line of. the transverse continuous tube (Ren Zijin (6)).

Some. information about 'the crack propagation at high nominal stresses was obtained from the fatigue resylts of the fabricated COD-specimens. For a K-value of 1440 N/mm3 2, da/dN was about

3.5 x 1'0 _{for cracks develo.ping in the longitudinal tube material}

and'about 2.5 X 10 '

for transverse cracks. These values are .about half as high as those for steel Fe 510..

BRITTLE FRACTURE 0F THE 'SPECIMEN AT 10°C

On page 3 the test procedure has been described. The. latter 1184 of. a total of 2284 load cyclgs had been applied at low temperature.

Of these 1068 occurred .at 5 C, the latter 116 at -10 C. The

fatigue load was +400 ton/-200 .ton, correspondiig to nominal stresses 0.27

a /0.13 a

y

The test ended with. a spontaneous. brittle', fracture,. Figures. 5

and 6 show the .crack paths oú bOth sides of the specimen. Figure 7 shows how far the brittle fracture extended in the transverse

tubes. This is astonishing bearing in mind that, 'in the greater'part of the transverse tubes hardly any. stresses were, present perpen-.dicular to the crack path. The starting point(s) of brittle

fracture. can 'be seen in Figures 8 and 9.. The fracture' probably: did

not., develop '.in one moment but in stages.. This can be 'seen in

Figure 8. It seems that at least four partial fractures.dëveioped prior to the final one. The strain gauge records confirm this more or less. Of six 'strain gauges a continuous record was 'made during the f atigue loading at low temperature. Unfortunately' they were not situated close.(enough) to the part of the structure shown in

Figure 8. But 'the. gauge near point A II-4 'in Figure. 8 was connec.ted to a peak detector (double amplitude of strain) which became active.

once. in '12-13 load cycles. The result is. shown in igure 1'0'.0(The

number of cycles refers to those effectuated at -5 C'and -'10 C). At. 1000. there is a clear change in slope which might be due to a

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

sudden extension of the shallow rusty fatigue crack at the edge of the plate. A second change of slope (to three horizontal points) occurs at about 1130 cycles. Next a small shift of the record is visible (two points), which again can only be caused by rather abrùpt crack extension. The starting point of the final fracture will, have been either at the left or the right side inFigure 8.

It is of interest that the partial brittle fractures occurred mainly in the welded region (weld and H.A.Z.) and.not in the base metal ofthe transverse tube. This confirms what will be discussed in the next section, that the resistance of the unwelded base material against.. longitudinal cracking was worse than that of the weld andH.A.Z.!

There is a possibility that primary cracking has started at the other side of the specimen (Figure 9), because the surface was more corroded than the one discussed before. Figure 11 which gives the record of the strain gauge in the vicinity of A i-6 in Figure 1.1, supports this view. The shear lips, all over the length of the fracture surf.ace.were rather small (Figure 8 left and right). Only at the origin and at a few intermediate spots and of course at the arresting points the shear lips were thicker.This. confirms that the propagation of the brittle fracture occurred very 'easily'. It suggests that the material of the transverse tube would probably. also not have been able to arrest cracks at somewhat higher tem-peratures, say 0°C. This is confirmed in the next section.

AGEING OF THE MATERIAL

The material had been supplièd in 1980. Then the Char.py energy at

-40°C was 44:J (notch J.. tube axis). Lloyd's requirement wás

32 J, so a wide margin was present.

After the experiment described in. this report new Charpy bars were fabricated and tested. (Figure 12).. Now only 32 J was.ob,tained.

at -40 C. it is curious that also the yield point and the tensile strength .had changed during the course of 5 years:

Iú order to be sure, that the same material was. used, the chemical composition was investigated. The result was identical to that of 1980.

Apparently this material. is sensitive to ageing. This would mean that structures made of that material (and perhaps similar ones) .are less safe nowadays than. at the time of fabrication.. This is important enough for justifying some additional testing..

161.6 1980 1985 a 810 885 y a u 860 910

FRACTURE CONTROL OF ENGINEER;ING STRUCTURES

_{- ECF 6}

Influence on Tensile Properties

First of all pieces of thesteel were subjected.to heating at 230 C

for two

hours. The -result was an increase, in yield point and

tensile strength of 20-30 N/mm2. Apparently .there was still room

for ageing after the '5 years' effect.

Next other pieces obtained a 'stress relief' treatment: 600°C for. two hours, in order to see whether the properties of 1980 could be recovered. The influence was ñegligible.

The next step was strain ageing consisting of 2.8% cold def or-mation followed by two, hours 'heating at 230 C'. This had a large

effect. The tensile strength became 130 N/mm2 higher than the 1985

'one, which' conforms to an increase of . 1,80

N/mm2' .as compared to 'the

1980 value. The elóngation was 'nearly halved.

Table 2 summarizes the results of the various treatments on the tensile properties.

TABLE 2

Specimen Treatment : a Elongation Reduction

No. ' " dp 100 ' of 'area 'TL - 2 none

90

. 885 19% 60% TL. - 3 aged 2h., 230°C 949 908 . 17% 63% TL - 4

"

"

938 '.905 17.5% 64% TL - 5' aged 2h.,600°C 928 888 19.4% ' 58%

TL -

6

strain .aged

7.9 (+2.. 8%)

52% 2.8%. 2h., 250 C

influence on Charpy

Values

The Charpy resulta were a bit surprising. Two hours ageing at 230 C led 'to a reduction in0'toughness in the low temperature region (Figures 15 and 16). At -10 C the toughness was improved in

_terms

.of energy (Figure 15) and. did not change in terms of fracture

appearance. But. on the 'whole it is evident 'that the toughness of the L-material has' been reduced to.that for the Dmaterial.

The results for specimens heated at 600°C are bad. The treat-ment. has certainly not restored the properties up'. to the 1980 level

as was-hOped for. This is not surprising 'as it0is known that for this' type of steel long lasting heating at '600 C maybe as well detrimental as beneficial. Therefore, although a positive result would have been of significance, a negative one -

as

found - says nothing.

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

The last treatment, being strain ageing, proves to be at least as detrimental as two hours heating at 230 C. But now the reduction

in toughness appears at -10 C, where heating alone had little effect.

DISCUSSION OF CHARPY AND COD VALUES IN CONNECTION TO INITIATION AND ARRESTING OF CRACKS

Charpy Tests

An alarming thing is that the Cha-rpy-value forbars with notch-es in the longitudinal direction was as low as 22 Jat -40 C. At -10°C (the testing temperature of the full scale specimen) the energy was still only 28 J. In Figure 12 are included requirements for this type of material according to Euronorm 113-72

(Fe E 355-KT). The correspondence is good.Then-it must be con-cluded that for offshore applications - especially for parts above sea level --these requirements are not satisfactory. Figure 13 -gives: information-about the fracture surface of the Charpy bars.

For the.tube thickness concernéd (20 mm) at least 5OZ shear surface will have to be required, when- crack arrest capability of the

-material is desired. But even for the L-bars this -will only be possible at about O C. From theD- andT-results, which are

-indicative for the full scale test result, it may be concluded that crack ar-resting will only be possible at temperatures above +20 C.

A point which needs consideration is whether the initiation of the brittle fracture was inflüenced or not by the quality of the weld and the heat-affected zone:. In Figure 14 Charpy results for weld and H.A.Z. òf a-similar structure as tested, made by the same firm, are added to the values .for the tube material of Figure 12. it may be concluded that the poor result of the full scale test was not due to inferior welding.

C.0.D. Testing

The C.O..D. tests showed a clear difference in-toughness -for cracking in the transverse and in the longitudinal direction of the tubes. For the transverse direction the CTOD at -169C was equal

-o o

to 0.7 mm and at -25 C and -40 C about 0.25 mm. For the longitudin-al direction it was only-O.22 mm at -4°C and 0.28 mm at +3°C. This result conforms to what has been observed in the- full scale test, where no brittle fractures hd developed in transverse sections of

the ]ongitudinal tube at -10 C.

-In none of the C.O.D. tests crack-arresting or pop-ins occurred. CONCLUSIONS -AND OBSERVATIONS

1. The high stres.s - low cycle -fatigue strength of as- welded tubular croas connections made of high -strength -steele is -about

equal to that for lower strength steels.

FRAcTURE CONTROL OF ENGINEERING

_{STRUCTURES -}

ECF 6

Shallow cracks developed into brittle fractures when fatigue loading was applied at moderately low temperatures.

These brittle fractures were not arrested when running in the longitudinal direction' of the tubes.

Ageing. of the material may aggravate the situation' for actual structures when they become older.

The notch toughness of tube material with respect to longitudin-al cracking is a lot worse than. that to transverse, cracking.

Fatigue loading at low temperature proved to be once again a very realistic and efficient procedure for testing large structural components.

It should be emphasized that the poor res,ults..of the experiment must be attributed 'entirely' to the bäd state of the tube material. The joint-design was good' as was:provenby the values of 'the

stresses'measured close to the initiation points of ,the cracks. REFERENCES

(1') Nibbering, J.J.'W. and Lalleman, A.W., 'Low Cycle Fatigue

'Tes'ts at LOW Temperature with E..G.-welded Plates', 1W-doc. X-593-70, 1970.

Nibber.ing:, J.J'..W.. and Scholte, H.G., 'Realistic Testing of

Welds 'by Fatigue Bending at Low Temperature',

11W-doc. X-1014-82, 1982'; I.S..P., Vol. 3.1, No. 355, March 1984.

Dijkstra, O.D., tFatigue Strength of, Tubular T- and X-joints'.,

OTC 3696, May 1980. '

De Back,, J., 'Vermoeiingsgedrag van Offshore Stalen Buis-knooppunten', Report for Annual Conférence of SMOZ, 1984. lida, K.,! 'Fatigue Strength: o'f Welded TubularK-joints of

8O0N/' Class HighStrength Steel', 11W-doc. XV-419-78,

1978. .

ReÙZijin, 'Fùl,1 Scale' Test andFinite Element Calcúlation for a Tubular X-joint', Graduate work Deif t 'University of Technology,, Ship Structures Laboratory, Deift, 19.85.

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

Crack Crack IillSlpprIIdsl FracIcr. Ilupparald.)

fr

A/8

_,/

_{Fratlur.I1.(und.rsiÔl}

/

rack W m.lSOmm altar Irautur. lundaruidi) Crack 11118mm iupp.ruidi)

As/B11

IFotncod.AI IFotncod,B)

161.10

/8

_-12

Figure 3. Situation of cracks (view from top side).

A

j'

Figure 7. ritt1e f raciires in transverse tube.

«f

--'

*'.Ç

FRACTURE CONTROL OF ENGINEERING STRUCTURES-

_{ECF 6}

I

161.11

I

Figure 2. Specimen in machine. Figure 5. Crack I (key Fig. 3).

Figure 2a. Detail. _{Figure 6. Crack IV (key Fig. 3).}

Si

4

Figure 8. Brittle fracture initia- _{Figure 9. Crack I (key Fig. 3).} tions and arrests. At the right

FRACTURE CONTROL OF ENGINEERING 'STRUCTURES

_{- ECF 6}

5500 5280 5000 3000 iooe

I

500 8rL. guide 32mm mue- 2 5.0.

Deift Univeruify of T.clmology/T.W.O.

t

Present fish high yield 860W/mm la' l.22.0. cycles i

tiitil_

i

ii iiiiil

1O 1O

Nümber of cycles (cAd of lest)

lai

Figure 4. Dutch test results on T- and X-joints of circular hollow sections' (3), (4).

situation of straisgauge

- I- i i i

0 200 400 600- 800 1000 1200 1400 1600 1800 2000

Number of Load cycles

Figure 10. Strain amplitüde close to A114 (Fig. 8).

161.12 ia8 u' T-0168 p.0.5 R0 A/!PB £ T-0457 .O,5 P.O An X - 0457 o. R.-I Au O T-0914 .O,5 8.0 Ax O X-0914 P.O5 8.0 Ax

FRACTURE CONTROL OF ENGINEERING STRUCTURES

- ECF 6

5-. o e situation of stroingaug I r i i r I I -200 400 600 800 1000 1200 1400 1600 1800. .2000

Nümber of toad ,cyc(ès

Figure 11. Strain amplitude near A1 6(Fig.. 9).

-._Le (iacn 113-fl. F.(3SS-KT Chupy-V USO-VI I 'I -10° 00 .100 Test temperature,(),

Figure 12-. Charpy-energy for tube material as compared to

Eurono-rm values. 16-1.13 80-70 Oand T.I...4 60

1

/,

L.3 .L6 3o -

,.

-Q.

-.

.130 'J

_Iii

_1I3 20 10 -0 I I I I -60° .500 -40° . -30° -20°

D.

o₁

.70

60

10

FRACTURE CONTROL OF ENGINEERING STRUCTURES

_{- ECF 6}

70 60 50 40 o 2O DJb II) 10 I4 _.-L54 ' -60 -5' -40 -30 .-20° -10° Test temperature 1°C)

Figure 13. Charpy fracture appearance.

112. 4H2. W2.I2. .wL WL

//

/

r

-40° -3 -20 -10 TAit temperatum (°C)

I

L8/

L. LI

).

/ L7

,

oW.H.

,

OW.It(2.) owt

/

,

--titor. lI,fl FrE 355-Kl cha.py-V lISO-VI t 16.1.14 t

Oand 1I....1

FigUre 14. Charpy-energy of -welds ánd H.A.Z. 's.

WNIwMd hLI flH. IHAI. hut) W2.lutdoiiildi) ud.) 90

:KJt

80

FRACTURE CONTROL OF ENGINEERING STRUCTURES

_{- ECF 6}

40 2 t-J 10 _o O -60 500 _{-40° -30°} 200 -io° 0° +10 Test temperature (PC)

Figure 15. Influence of various ageing treatments in terms of Charpy-energy.

/

/

/

'o

o O D 2hour23O°( O 2Puri6OOCt A 2/.iIraAt.Imu,,.2UOt AU potnt. L-FAIA,411 o w a o

,

,,

,

8 -50° 600

-o°

_-20° 10C) Test temperature (°() 161.15 o o 0° uP

Figure 16. Influence of various ageing treatments in terms of fracture appearance.

70 60 50 w 'o 40 Q llojrs'230 ct O 2Io.ir6OO 28%.tran.Utw. 2501t AU poiLmateriaI

/

/

/

/

/

D

,

,

/

A

. /

70-60 50